Responses of clonal architecture to experimental defoliationa comparative study between ten grassland species

Marie-Lise Benot AElig Cendrine Mony AElig Sara Puijalon AEligMajid Mohammad-Esmaeili AElig Jacques J M van Alphen AEligJan-Bernard Bouzille AElig Anne Bonis

Received 19 June 2008 Accepted 24 October 2008

Springer Science+Business Media BV 2008

Abstract Clonal architecture may enable plants to

effectively respond to environmental constraints but

its role in plant tolerance to defoliation remains

poorly documented In several non-clonal species

modifications of plant architecture have been

reported as a mechanism of plant tolerance to

defoliation yet this has been little studied in clonal

plants In a glasshouse experiment five rhizomatous

and five stoloniferous species of grazed pastures were

subjected to three frequencies of defoliation in order

to test two hypotheses (1) We expected plant clonal

response to defoliation to be either a more compact

architecture (low clonal propagation but high

branching) or a more dispersed one (long-distance

propagation and low branching) Such plastic adjust-

ments of clonal architecture were assumed to be

involved in tolerance to defoliation ie to promote

genet performance in terms of biomass and number of

ramets (2) The response of clonal architecture to

defoliation was expected to be dependent on the

species and to be more plastic in stoloniferous than in

rhizomatous species Most genets of each species

were tolerant to defoliation as they survived and

developed in every treatment Architectural modifi-

cations in response to defoliation did not match our

predictions Clonal growth was either maintained or

reduced under defoliation Relative growth rate

(RGR) decreased in eight species whereas defoliated

genets of seven species produced as many ramets as

control genets Biomass allocation to ramet shoots

remained stable for all but one species In defoliated

genets the number and mean length of connections

and mean inter-ramet distance were equal to or lower

than those in control genets Four groups of species

were distinguished according to their architectural

response to defoliation and did not depend on the

type of connections We hypothesised that dense

clonal architectures with low plasticity may be the

most advantageous response in defoliated conditions

such as in grazed pastures

Keywords Between-class PCA Glasshouse experiment Response groups Short-term defoliation Tolerance

Electronic supplementary material The online version ofthis article (doi101007s11258-008-9546-3) containssupplementary material which is available to authorized users

M-L Benot (amp) C Mony M Mohammad-Esmaeili J J M van Alphen J-B Bouzille A Bonis

UMR CNRS 6553 ECOBIO Universite de Rennes 1

Campus de Beaulieu 35042 Rennes Cedex France

e-mail marie-lisebenotuniv-rennes1fr

S Puijalon

UMR CNRS 5023 Ecology of Fluvial Hydrosystems

Universite Lyon 1 69622 Villeurbanne Cedex France

M Mohammad-Esmaeili

Faculty of Rangeland and Watershed Gorgan University

of Agricultural Sciences and Natural Resources Shahid

Beheshti Street Gorgan Iran

123

Plant Ecol

DOI 101007s11258-008-9546-3

Introduction

Vegetative multiplication is widely spread in plants

particularly in Angiosperms Many species are able to

reproduce both by seeds (sexual reproduction) and by

clonal growth (asexual reproduction) while some

others are even exclusively clonal (Price and

Marshall 1999) A genet consists of the vegetative

production of genetically identical offsprings (ra-

mets) that are potentially independent units as they

develop their own shoots and roots (Hutchings and

Bradbury 1986 Klimes et al 1997) This propagation

allows the genet to persist and spread both in space

and time (Gardner and Mangel 1997 Oborny and

Kun 2002) Vegetative growth modes are variable

and major attention has been paid to clonal plants

forming aboveground or belowground horizontal

stems (stolons and rhizomes respectively hereafter

often referred as connections) carrying ramets

(Klimes et al 1997) In the following the term

clonality will refer to these particular growth modes

Clonal architecture provides singular plant char-

acteristics relying mainly on the integration between

ramets which presents a potential adaptive role

(Hutchings and Wijesinghe 1997 Suzuki and Stuefer

1999) Clonal plants are likely to effectively respond

to environmental constraints that may explain their

abundance in a variety of environments (Hutchings

1999 Price and Marshall 1999) Clonal plants are

particularly able to show plastic adjustments of clonal

architecture a strategy involved in selective foraging

for light quality (Stuefer and Huber 1998) light

intensity (Dong and Pierdominici 1995) nutrient

availability (Liao et al 2003) and even soil temper-

ature (MacDonald and Lieffers 1993) and

competition (MacDonald and Lieffers 1993 Kleijn

and van Groenendael 1999 Macek and Leps 2003)

For plants submitted to defoliation the ability of

growth and reproduction after damage is defined as

tolerance while the term compensation is often used

to characterise the degree of this tolerance (Strauss

and Agrawal 1999) Tolerance can be considered as a

plastic trait and ranged along a gradient (Maschinski

and Whitham 1989 Stowe et al 2000) Incomplete

tolerance occurs when defoliated plants survive and

develop but their performance is lower than for

undefoliated plants (undercompensation sensu

Strauss and Agrawal 1999) Compensating and even

overcompensating tolerance respectively refer to

maintained and increased performance for damaged

plants compared to undamaged ones (Stowe et al

2000) Amongst a variety of mechanisms modifica-

tions of aboveground plant architecture (ie

branching pattern) had often been mentioned as a

frequent response to clipping in non-clonal plants

(Owen 1980 Paige and Whitham 1987 Lennartsson

et al 1998)

The aim of this study was to investigate whether

active adjustments of clonal architecture are involved

in genet tolerance to defoliation Few studies have

been carried out on the responses of clonal architec-

ture to disturbance such as clipping or grazing (see

however Moen et al 1999 Piqueras 1999 Li et al

2004 Wang et al 2004) We investigated such

questions by considering the comparative response

to experimental clipping for 10 species embracing

both stoloniferous and rhizomatous species

We first tested whether clonal architecture-related

traits are involved in species compensating tolerance

of defoliation enabling the maintenance of genet

performance (in terms of biomass and ramet

production) Their response is expected to vary

between traits and either a more compact architec-

ture (low clonal propagation but high branching) or

a more dispersed one (long-distance propagation and

low branching) is expected to occur in response to

defoliation (Table 1) The second hypothesis was

that the response of clonal architecture to defoliation

might vary according to species and in particular

Table 1 Variations of clonal traits predicted by the hypothesis

of compensating tolerance to defoliation

Survival and development

Clonal performance

Relative growth rate

Number of ramets

Biomass allocation to ramets

Clonal architecture-related traits

Number of connections Mean length of connections Mean inter-ramet distance

Clonal growth form Dispersed Compact

Arrows indicate the direction (decrease maintenance or

increase) of trait variation between undefoliated and

defoliated plants

Plant Ecol

123

among stoloniferous and rhizomatous species As

stolons are more often involved in spatial propaga-

tion and show higher morphological plasticity than

rhizomes (Dong and de Kroon 1994 Dong and

Pierdominici 1995) clonal architecture is expected

to be more responsive to defoliation in stoloniferous

than in rhizomatous species

Methods

The response of 10 clonal species to three frequencies

of defoliation was recorded in terms of genet

performance and architecture-related traits

The plant species for study were selected out of

the 23 clonal herbaceous perennials from grazed

pastures in the Marais poitevin Western France

(46280 N and 1300 W) These species are the most

abundant clonal species in these plant communities

where the biomass consumption by grazing ranges

from 55 to 87 of the available biomass (Ros-

signol et al unpublished) They belong to several

families and show different modes of clonal growth

All species can produce long connections either

aboveground (stolons) or belowground (rhizomes)

or both Some of them can also form tussocks

through very short connections (caespitose growth

form) (Table 2)

Experimental set-up

Ramets were collected in February 2006 from a

grazed area of 1 ha Thirty-three ramets per species

were chosen and randomly assigned to one of the

three treatments of defoliation The experimental

design thus consisted of 10 species 9 3 defoliation

treatments 9 11 replicates with a total of 330 exper-

imental units

Each ramet was cleaned weighed (fresh mass

FM) and rooted in the centre of a pot (20-cm-

diameter and 16-cm-height) containing fine garden

soil Approximately 6ndash10 cm3 of substrate from the

collection site were added close to the roots of the

ramet in order to provide symbiotic microorganisms

Ramets were first allowed to grow freely for a five-

week acclimation period during which dead ramets

were replaced The ratio FMDM (fresh and dry mass

respectively) was determined for 10 additional non-

planted ramets per species and the mean per species

used to estimate initial dry biomass of each planted

ramet

The experiment was conducted in a non-heated

glasshouse at the campus of Beaulieu (University of

Rennes 1 France) from 29th March up to 17th May

2006 In the glasshouse temperature was maintained

below 25C Pots were watered daily with tap water

and weeds were regularly removed

Table 2 Studied species and their clonal growth type (adapted from Klimes et al 1997)

Species Abbreviation Class Family Clonal growth type

Agrostis stolonifera Asto Monocotyledons Poaceae Stoloniferous

Caespitose

Carex divisa Cdiv Monocotyledons Cyperaceae Rhizomatous

Eleocharis palustris Epal Monocotyledons Cyperaceae Rhizomatous

Elytrigia repens Erep Monocotyledons Poaceae Caespitose

Rhizomatous

Glyceria fluitans Gflu Monocotyledons Poaceae Caespitose

Stoloniferous

Juncus articulatus Jart Monocotyledons Juncaceae Rhizomatous

Stoloniferous

Juncus gerardii Jger Monocotyledons Juncaceae Rhizomatous

Ranunculus repens Rrep Dicotyledons Ranunculaceae Stoloniferous

Trifolium fragiferum Tfra Dicotyledons Fabaceae Stoloniferous

Trifolium repens Trep Dicotyledons Fabaceae Stoloniferous

For species having the ability to develop two types of connections the major type developed during the experiment is mentioned first

Caespitose growth type corresponds to the production of short rhizomes (tussock forming species)

Plant Ecol

123

Experimental treatments

The three frequencies of defoliation tested were no

defoliation (control treatment) defoliation every

30 days (mid-frequency defoliation treatment) and

defoliation every 15 days (high-frequency defoliation

treatment) To make a realistic simulation of defoli-

ation caused by cattle all aboveground parts of

Monocotyledons were cut to 7 cm height (Loucou-

garay et al 2004) As Dicotyledons were generally

lower than 7 cm defoliation consisted in the removal

of 50 of developed leaves by severing the petiole at

its base (both petiole and lamina were removed)

Genets were harvested 8 weeks after the first clip-

ping Genets under mid-frequency treatment were

thus cut twice and those under high-frequency

treatment were cut four times

Trait measurements

At harvest the number of ramets per genet was

counted and architectural traits were measured Then

each genet was divided into vegetative shoots flow-

ers connections and roots dried to constant mass at

60C and the dry mass of each of these parts was

weighed As the study focused on clonal architecture

biomass allocation to roots was not taken into account

As only a few genets had produced flowers during the

experiment only traits related to vegetative develop-

ment were taken into account

Traits related to clonal performance

The investment in clonal reproduction was estimated

as the final number of ramets The total growth of

each genet was measured as the relative growth rate

calculated as follows

RGR

frac14 lnethfinal biomassthorn clippingsTHORN lnethinitial biomassTHORNnumber of days

where final biomass is the dry mass of the whole

genet (including roots) at the end of the experiment

clippings the dry mass of clipped tissues and initial

biomass the dry mass of the planted ramet

Finally the biomass of ramets corresponded to the

sum of the final dry mass of shoots of all ramets

produced by the genet

Architecture-related traits

Measured traits were the total number of connections

(stolons or rhizomes) per genet the mean length of

connections (with a precision of 1 mm) produced per

genet and the mean inter-ramet distance (with a

precision of 1 mm) per genet This latter corre-

sponded to the mean distance between two

consecutive ramets It could be calculated only for

genets that had produced connections carrying ramets

(Fig 1)

Four out of 10 species can develop two types of

connections (Table 2) However Juncus articulatus

produced very few stolons and only data related to

rhizomes were taken into account For Agrostis

stolonifera Elytrigia repens and Glyceria fluitans

the calculation of mean inter-ramet distance did not

include caespitose connections

A species was considered as tolerant when genets

had survived and developed even under the high-

frequency defoliation treatment The response of clonal

performance to defoliation (from a decrease to an

increase) were used to characterise the degree of

tolerance (from incomplete to compensating tolerance)

Statistical analysis

For all species one-way ANOVAs showed no

differences in ramet initial biomass between the

three treatments confirming the assumption of ramet

randomisation between treatments at the beginning of

the experiment The percentage of biomass removed

by defoliation was assessed through ANOVA with

species and treatment as main effects The correlation

between the percentage of biomass removed and the

ramet

connection length

inter-ramet distance

shoots

roots

connection

ramet

connection length

inter-ramet distance

shoots

roots

connection

Fig 1 Schematic representation of a genet and definition of

architectural traits

Plant Ecol

123

values of architecture-related traits was tested We

used the non-parametric Spearman correlation coef-

ficient (q) as traits did not follow a normal

distribution

Within-species effects of defoliation treatments on

final number of ramets RGR mean length of

connections and mean inter-ramet distance were

assessed through one-way ANOVAs after checking

for normality and homogeneity of variances (Kol-

mogorovndashSmirnov and Levene tests respectively)

and log-transformation of data when necessary Post-

hoc comparisons between treatments were tested by

the TukeyndashKramer test In the particular case of final

number of connections for which assumptions of

normality and homogeneity of variances were not

met non-parametric KruskalndashWallis tests were used

and post-hoc comparisons were made by Mann and

Whitney U-tests with Bonferroni correction The

effect of defoliation on biomass allocation to ramets

was analysed by ANCOVAs using final biomass as a

covariate The aim was to increase the power of the

F-statistic by adjusting for the influence of the

covariate and to avoid the use of biomass ratios

which may be misleading to study allocation patterns

(Jiasenski and Bazzaz 1999) Interactions between

treatment and covariate were first introduced into the

model and removed when non-significant

The comparison of architectural responses to

defoliation between the 10 species was done by

multivariate analyses taking into account the three

architecture-related traits After a Principal Compo-

nents Analysis (PCA) on correlation matrix traits

were centred independently per species (within-

species PCA) and compared between treatments by

a between-class PCA (bc-PCA) each treatment

considered as one class (Doledec and Chessel

1991) Such analysis enabled to eliminate scale

effects linked to differences of average trait values

between species Following this analysis hierarchical

ascendant classification (HAC using Ward method of

clustering Legendre and Legendre 1998) was used to

group species according to their multivariate trajec-

tory of response The coordinates of each treatment

along the two first axes of bc-PCA constituted the six

variables

ADE-4 software (Thioulouse et al 1997) was used

for bc-PCA and JMP software (SAS procedure) for

other statistical calculations In all the cases we

rejected null hypothesis at the P = 005 level

Results

Effects of defoliation treatments on clonal

performance and architecture-related traits

The amount of biomass removed by defoliation

differed significantly between species and treatments

(species 9 treatment F18284 frac14 1205 P 00001

species F9284 frac14 3678 P 00001 treatment F2

284 frac1475004 P 00001) For the high-frequency defoli-

ation treatment it ranged from 425 (Agrostis

stolonifera) to 10 (Juncus gerardii) The three

grasses were the most severely impacted (between

27 and 425 for high-frequency defoliation and

between 20 and 24 for mid-frequency treatment)

whereas J gerardii and the two Trifolium species

were less impacted especially by mid-frequency

defoliation treatment (only from 5 to 7 biomass

removed) (Fig 2)

At the end of the experiment 311 out of the 330

genets had survived and developed all species and

treatments taken together The impact of treatments

on trait values depended on species and traits

Significant differences of trait values occurred either

between the control and both defoliation treatments

or between the control and the high-frequency

defoliation treatment only Therefore hereafter the

term defoliation will most frequently be used without

distinction between the two levels of the defoliation

treatment J articulatus J gerardii and Ranunculus

repens were the three species for which almost no

trait was significantly impacted by defoliation By

contrast the final number of ramets of defoliated

genets was 60 to 30 compared to the one of

control genets for Eleocharis palustris G fluitans

and Trifolium fragiferum It did not significantly

change for the other ones (Table 3 see also Table S1-

A in Supplementary material) Defoliation generated

a significant reduction of RGR for eight species

(Table 3 Table S1-B) and there was a significant

covariation between the final biomass and the

biomass of ramet shoots for all species except T

fragiferum This last trait was significantly impacted

by clipping treatment only for Trifolium repens

indicating that for the nine other species biomass

allocation to shoots (ie the part of the genet final

biomass allocated to the ramet shoots) was not

changed by defoliation (Table 3 Table S1-C) For

Carex divisa only all architecture-related traits

Plant Ecol

123

Experimental treatments

The three frequencies of defoliation tested were no

defoliation (control treatment) defoliation every

30 days (mid-frequency defoliation treatment) and

defoliation every 15 days (high-frequency defoliation

treatment) To make a realistic simulation of defoli-

ation caused by cattle all aboveground parts of

Monocotyledons were cut to 7 cm height (Loucou-

garay et al 2004) As Dicotyledons were generally

lower than 7 cm defoliation consisted in the removal

of 50 of developed leaves by severing the petiole at

its base (both petiole and lamina were removed)

Genets were harvested 8 weeks after the first clip-

ping Genets under mid-frequency treatment were

thus cut twice and those under high-frequency

treatment were cut four times

Trait measurements

At harvest the number of ramets per genet was

counted and architectural traits were measured Then

each genet was divided into vegetative shoots flow-

ers connections and roots dried to constant mass at

60C and the dry mass of each of these parts was

weighed As the study focused on clonal architecture

biomass allocation to roots was not taken into account

As only a few genets had produced flowers during the

experiment only traits related to vegetative develop-

ment were taken into account

Traits related to clonal performance

The investment in clonal reproduction was estimated

as the final number of ramets The total growth of

each genet was measured as the relative growth rate

calculated as follows

RGR

frac14 lnethfinal biomassthorn clippingsTHORN lnethinitial biomassTHORNnumber of days

where final biomass is the dry mass of the whole

genet (including roots) at the end of the experiment

clippings the dry mass of clipped tissues and initial

biomass the dry mass of the planted ramet

Finally the biomass of ramets corresponded to the

sum of the final dry mass of shoots of all ramets

produced by the genet

Architecture-related traits

Measured traits were the total number of connections

(stolons or rhizomes) per genet the mean length of

connections (with a precision of 1 mm) produced per

genet and the mean inter-ramet distance (with a

precision of 1 mm) per genet This latter corre-

sponded to the mean distance between two

consecutive ramets It could be calculated only for

genets that had produced connections carrying ramets

(Fig 1)

Four out of 10 species can develop two types of

connections (Table 2) However Juncus articulatus

produced very few stolons and only data related to

rhizomes were taken into account For Agrostis

stolonifera Elytrigia repens and Glyceria fluitans

the calculation of mean inter-ramet distance did not

include caespitose connections

A species was considered as tolerant when genets

had survived and developed even under the high-

frequency defoliation treatment The response of clonal

performance to defoliation (from a decrease to an

increase) were used to characterise the degree of

tolerance (from incomplete to compensating tolerance)

Statistical analysis

For all species one-way ANOVAs showed no

differences in ramet initial biomass between the

three treatments confirming the assumption of ramet

randomisation between treatments at the beginning of

the experiment The percentage of biomass removed

by defoliation was assessed through ANOVA with

species and treatment as main effects The correlation

between the percentage of biomass removed and the

ramet

connection length

inter-ramet distance

shoots

roots

connection

ramet

connection length

inter-ramet distance

shoots

roots

connection

Fig 1 Schematic representation of a genet and definition of

architectural traits

Plant Ecol

123

values of architecture-related traits was tested We

used the non-parametric Spearman correlation coef-

ficient (q) as traits did not follow a normal

distribution

Within-species effects of defoliation treatments on

final number of ramets RGR mean length of

connections and mean inter-ramet distance were

assessed through one-way ANOVAs after checking

for normality and homogeneity of variances (Kol-

mogorovndashSmirnov and Levene tests respectively)

and log-transformation of data when necessary Post-

hoc comparisons between treatments were tested by

the TukeyndashKramer test In the particular case of final

number of connections for which assumptions of

normality and homogeneity of variances were not

met non-parametric KruskalndashWallis tests were used

and post-hoc comparisons were made by Mann and

Whitney U-tests with Bonferroni correction The

effect of defoliation on biomass allocation to ramets

was analysed by ANCOVAs using final biomass as a

covariate The aim was to increase the power of the

F-statistic by adjusting for the influence of the

covariate and to avoid the use of biomass ratios

which may be misleading to study allocation patterns

(Jiasenski and Bazzaz 1999) Interactions between

treatment and covariate were first introduced into the

model and removed when non-significant

The comparison of architectural responses to

defoliation between the 10 species was done by

multivariate analyses taking into account the three

architecture-related traits After a Principal Compo-

nents Analysis (PCA) on correlation matrix traits

were centred independently per species (within-

species PCA) and compared between treatments by

a between-class PCA (bc-PCA) each treatment

considered as one class (Doledec and Chessel

1991) Such analysis enabled to eliminate scale

effects linked to differences of average trait values

between species Following this analysis hierarchical

ascendant classification (HAC using Ward method of

clustering Legendre and Legendre 1998) was used to

group species according to their multivariate trajec-

tory of response The coordinates of each treatment

along the two first axes of bc-PCA constituted the six

variables

ADE-4 software (Thioulouse et al 1997) was used

for bc-PCA and JMP software (SAS procedure) for

other statistical calculations In all the cases we

rejected null hypothesis at the P = 005 level

Results

Effects of defoliation treatments on clonal

performance and architecture-related traits

The amount of biomass removed by defoliation

differed significantly between species and treatments

(species 9 treatment F18284 frac14 1205 P 00001

species F9284 frac14 3678 P 00001 treatment F2

284 frac1475004 P 00001) For the high-frequency defoli-

ation treatment it ranged from 425 (Agrostis

stolonifera) to 10 (Juncus gerardii) The three

grasses were the most severely impacted (between

27 and 425 for high-frequency defoliation and

between 20 and 24 for mid-frequency treatment)

whereas J gerardii and the two Trifolium species

were less impacted especially by mid-frequency

defoliation treatment (only from 5 to 7 biomass

removed) (Fig 2)

At the end of the experiment 311 out of the 330

genets had survived and developed all species and

treatments taken together The impact of treatments

on trait values depended on species and traits

Significant differences of trait values occurred either

between the control and both defoliation treatments

or between the control and the high-frequency

defoliation treatment only Therefore hereafter the

term defoliation will most frequently be used without

distinction between the two levels of the defoliation

treatment J articulatus J gerardii and Ranunculus

repens were the three species for which almost no

trait was significantly impacted by defoliation By

contrast the final number of ramets of defoliated

genets was 60 to 30 compared to the one of

control genets for Eleocharis palustris G fluitans

and Trifolium fragiferum It did not significantly

change for the other ones (Table 3 see also Table S1-

A in Supplementary material) Defoliation generated

a significant reduction of RGR for eight species

(Table 3 Table S1-B) and there was a significant

covariation between the final biomass and the

biomass of ramet shoots for all species except T

fragiferum This last trait was significantly impacted

by clipping treatment only for Trifolium repens

indicating that for the nine other species biomass

allocation to shoots (ie the part of the genet final

biomass allocated to the ramet shoots) was not

changed by defoliation (Table 3 Table S1-C) For

Carex divisa only all architecture-related traits

Plant Ecol

123

values of architecture-related traits was tested We

used the non-parametric Spearman correlation coef-

ficient (q) as traits did not follow a normal

distribution

Within-species effects of defoliation treatments on

final number of ramets RGR mean length of

connections and mean inter-ramet distance were

assessed through one-way ANOVAs after checking

for normality and homogeneity of variances (Kol-

mogorovndashSmirnov and Levene tests respectively)

and log-transformation of data when necessary Post-

hoc comparisons between treatments were tested by

the TukeyndashKramer test In the particular case of final

number of connections for which assumptions of

normality and homogeneity of variances were not

met non-parametric KruskalndashWallis tests were used

and post-hoc comparisons were made by Mann and

Whitney U-tests with Bonferroni correction The

effect of defoliation on biomass allocation to ramets

was analysed by ANCOVAs using final biomass as a

covariate The aim was to increase the power of the

F-statistic by adjusting for the influence of the

covariate and to avoid the use of biomass ratios

which may be misleading to study allocation patterns

(Jiasenski and Bazzaz 1999) Interactions between

treatment and covariate were first introduced into the

model and removed when non-significant

The comparison of architectural responses to

defoliation between the 10 species was done by

multivariate analyses taking into account the three

architecture-related traits After a Principal Compo-

nents Analysis (PCA) on correlation matrix traits

were centred independently per species (within-

species PCA) and compared between treatments by

a between-class PCA (bc-PCA) each treatment

considered as one class (Doledec and Chessel

1991) Such analysis enabled to eliminate scale

effects linked to differences of average trait values

between species Following this analysis hierarchical

ascendant classification (HAC using Ward method of

clustering Legendre and Legendre 1998) was used to

group species according to their multivariate trajec-

tory of response The coordinates of each treatment

along the two first axes of bc-PCA constituted the six

variables

ADE-4 software (Thioulouse et al 1997) was used

for bc-PCA and JMP software (SAS procedure) for

other statistical calculations In all the cases we

rejected null hypothesis at the P = 005 level

Results

Effects of defoliation treatments on clonal

performance and architecture-related traits

The amount of biomass removed by defoliation

differed significantly between species and treatments

(species 9 treatment F18284 frac14 1205 P 00001

species F9284 frac14 3678 P 00001 treatment F2

284 frac1475004 P 00001) For the high-frequency defoli-

ation treatment it ranged from 425 (Agrostis

stolonifera) to 10 (Juncus gerardii) The three

grasses were the most severely impacted (between

27 and 425 for high-frequency defoliation and

between 20 and 24 for mid-frequency treatment)

whereas J gerardii and the two Trifolium species

were less impacted especially by mid-frequency

defoliation treatment (only from 5 to 7 biomass

removed) (Fig 2)

At the end of the experiment 311 out of the 330

genets had survived and developed all species and

treatments taken together The impact of treatments

on trait values depended on species and traits

Significant differences of trait values occurred either

between the control and both defoliation treatments

or between the control and the high-frequency

defoliation treatment only Therefore hereafter the

term defoliation will most frequently be used without

distinction between the two levels of the defoliation

treatment J articulatus J gerardii and Ranunculus

repens were the three species for which almost no

trait was significantly impacted by defoliation By

contrast the final number of ramets of defoliated

genets was 60 to 30 compared to the one of

control genets for Eleocharis palustris G fluitans

and Trifolium fragiferum It did not significantly

change for the other ones (Table 3 see also Table S1-

A in Supplementary material) Defoliation generated

a significant reduction of RGR for eight species

(Table 3 Table S1-B) and there was a significant

covariation between the final biomass and the

biomass of ramet shoots for all species except T

fragiferum This last trait was significantly impacted

by clipping treatment only for Trifolium repens

indicating that for the nine other species biomass

allocation to shoots (ie the part of the genet final

biomass allocated to the ramet shoots) was not

changed by defoliation (Table 3 Table S1-C) For

Carex divisa only all architecture-related traits

Plant Ecol

123

showed a significant drop after defoliation (Table 3)

For E palustris E repens and T fragiferum the

number of connections decreased up to 80 for E

palustris (Table 3 Table S2-A) Mean length of

connections decreased by 75 for A stolonifera and

50 for C divisa and G fluitans (Table 3 Table S2-

B) Finally mean inter-ramet distance decreased in A

stolonifera C divisa and T repens (Table 3 Table

S2-C)

Multivariate responses of clonal architecture to

defoliation

Over all 10 species there was no significant correla-

tion between the percentage of biomass removed and

the three architectural traits (q = -008 P = 015 for

the number of connections q = 002 P = 068 for

the mean length of connections and q = 002

P = 068 for the mean inter-ramet distance) But

their response to defoliation varied between traits and

species The two first axes of bc-PCA represented

96 of total variance (69 and 27 respectively

Fig 3) The F1 axis carried out mean length of

connections and mean inter-ramet distance The F2

axis was negatively correlated with the number of

connections (Fig 3a) The amplitude of variation

between the extremes of the trajectories along the first

axis was weak for all species but A stolonifera The

trajectory between control and high-frequency defo-

liation treatment along the second axis varied in the

direction of a decreased number of connections for

nine species This variation was the most important

)(

ssamoib

devomer

High-frequency defoliation treatment

Mid-frequency defoliation treatment

0

10

20

30

40

50 a

b

bc bcdbcd cde

cde

de e

fgh

high

gh

hihi

ef

cdede

efgde

Asto Erep Rrep Gflu Cdiv Tfra Jart Trep Epal Jger

Fig 2 Means and standard

deviations of percentage of

biomass removed [biomass

removed(final

biomass biomass

removed)] for each species

Letters indicate significant

differences between

treatments and species

(post-hoc Tukey tests)

Table 3 Effects of defoliation on clonal traits for all the 10 species

First hypothesis Asto Cdiv Epal Erep Gflu Jart Jger Rrep Tfra Trep

Survival and development

Clonal performance

RGR Number of ramets

Biomass allocation to rametsa

Clonal architecture

Number of connections

Mean length of connections

Mean inter-ramet distance NA NA

Arrows indicate the variations of trait values between control and both defoliation treatments taken altogether significant increase

significant decrease no significant difference NA not available Abbreviations of species follow Table 2 Results of statistical

tests are presented in Supplementary material (Table S1 for traits indicative of performance Table S2 for architectural traits)a Biomass allocation to ramets corresponds to the effect of treatment (main factor) on the biomass of all ramets of a genet tested by

the ANCOVA (final genet biomass as a covariate)

Plant Ecol

123

for Trifolium species weaker for C divisa E palus-

tris E repens and G fluitans and was close to zero

for Juncaceae and R repens The trajectory of

A stolonifera along this axis had a singular shape

with a great increase between control and mid-

frequency defoliation and a decrease between mid-

frequency and high-frequency defoliation (Fig 3b)

The HAC based on architectural responses to

defoliation resulted in four groups of species The

first group consisted of both Trifolium species

J articulatus J gerardii and R repens formed the

second group while C divisa E palustris E repens

and G fluitans constituted the third group Finally the

last group corresponded to A stolonifera due to the

singular shape of its trajectory of response (Fig 4)

Discussion

Consequences of defoliation on clonal

performance

All species showed tolerance to defoliation as 311 out

of 330 genets survived and developed even when

severely clipped

Biomass allocation to ramets was not affected by

the treatment except for one species This indicates

that genets were able to sustain damaged ramets and

even to compensate for aboveground biomass

removal caused by defoliation While the design did

not allow to identify the mechanisms involved

B

Tfra Trep

Asto

Rrep

ErepEpal

Gflu

C

M

H

C

M

H

C

M

H

CMH

CM

H

C

M

H

C

M

H

C

M

HC

M

H

C

M

H

Cdiv

Jart Jger

NbCon

11 -1

-1

A

72 2F

-037-11 13

046Tfra Trep

Asto

Rrep

ErepEpal

Gflu

C

M

H

C

M

H

C

M

H

CMH

CM

H

C

M

H

C

M

H

C

M

HC

M

H

C

M

H

Cdiv

Jart Jger

NbCon

LConDist

11 -1

-1

F1 69

-037-11 13

046

Fig 3 Multivariate responses of clonal architecture to defo-

liation for each species Traits values have been centred per

species (within-species PCA) and compared between treat-

ments (between-treatment PCA) The factorial plan is thus the

same for all species a Correlation circle of architectural traits

in the factorial plan 1ndash2 of between-treatment PCA bTrajectories of multivariate responses to the three treatments

for each species in the factorial plan 1ndash2 Each point represents

the barycentre of all replicates of a defoliation treatment

Points C Control treatment M mid-frequency defoliation

treatment H high-frequency defoliation treatment Dist mean

inter-ramet distance LCon mean length of connections

NbCon number of connections See Table 2 for the meaning

of species abbreviations

Tfra

Trep

Rrep

Gflu

Asto

Cdiv

Epal

Erep

Jart

Jger

Fig 4 Dendrogram of species resulting from HAC according

to the multivariate responses of clonal architecture to

defoliation Species were clustered into four groups on the

basis of Euclidean distance (Wardrsquos method) See Table 2 for

the meaning of species abbreviations

Plant Ecol

123

Brown and Allen (1989) reported that clipping

treatment may cause the translocation of resources

from belowground organs to support the regrowth of

aboveground tissues

We recorded however a great discrepancy on the

degree of tolerance depending on traits and species

considered The RGR decreased with clipping which

is typical of undercompensation (Strauss and Agra-

wal 1999) and incomplete tolerance (Stowe et al

2000) while together seven out of 10 species

maintained a similar number of ramets in all three

treatments arguing for compensating tolerance to

clipping Previous studies had already reported a

variety of responses to clipping by clonal plants

among species and among environments for a given

species For example ramet number has been

reported to decrease with leaf removal for three

clonal perennials (Hicks and Turkington 2000) to be

unchanged for Leymus chinensis (Wang et al 2004)

either to be unchanged or to increase according to

nutrient availability for Cyperus esculentus (Li et al

2004) and even to increase for five caespitose grasses

(Richards et al 1988) Biomass responses to

defoliation were also shown to vary from undercom-

pensation (Li et al 2004) to overcompensation

(Hicks and Turkington 2000) probably due to both

the differing species studied and the defoliation

treatment applied Interspecific differences in com-

pensatory ability were notably found both in clonal

and non-clonal species (McNaughton and Chapin

1985 Belsky 1986 Del-Val and Crawley 2005)

Response to defoliation has also been shown to vary

within the same species depending on environmental

conditions (Maschinski and Whitham 1989) The lack

of generality in clonal plant responses to clipping

may also originate from the variety of defoliation

treatments used in the different studies Their impact

on plant growth may indeed differ whether they are

applied at a given date (eg Hicks and Turkington

2000 Wang et al 2004) at a given development

stage (eg Richards et al 1988) or several times (eg

Li et al 2004)

Responses of clonal architecture-related traits

to defoliation

Clonal architecture-related traits did not match the

predicted responses Species followed four types of

architectural response to defoliation going from nil

to high variation Trait values never increased with

clipping and defoliation thus led to fewer connections

andor shorter connections and mean inter-ramet

distances Similar results in response to clipping or

grazing have already been observed in other herba-

ceous clonal plants such as Trifolium repens (Hay and

Newton 1996) Acaena magellanica (Moen et al

1999) Trientalis europaea (Piqueras 1999) or Lymus

chinensis (Wang et al 2004)

Furthermore the results obtained here show that

there is no relationship between the multivariate

pattern of architectural response and the degree of

tolerance to defoliation For instance both the first

and the third response groups included together

species for which defoliation induced no change in

the number of ramets (C divisa E repens and T

repens) and species for which defoliation induced a

decreased number of ramets (E palustris G fluitans

and T fragiferum) Compensating tolerance can be

related to various responses of architectural traits For

E repens the production of ramets was not affected

by defoliation despite a decrease of the number of

connections as only a few ramets were produced by

these connections the majority being caespitose

(tussock forming) An alternative strategy was shown

by C divisa T repens and A stolonifera They

maintained the number of ramets unchanged with

clipping by way of the decreased mean inter-ramet

distance whatever the variation of the other traits

Interspecific comparison of the responses

of clonal architecture to defoliation

Previous studies have shown that higher nutrient and

or light supply increased branching intensities of both

stolons and rhizomes whereas the morphology of

stolons (mean connection length and mean internode

length) was more plastic than the morphology of

rhizomes (Dong and de Kroon 1994 Dong and

Pierdominici 1995) According to our study only the

stolons of A stolonifera showed a high degree of

variability with a high amplitude of variation of all

architecture-related traits The responses of other

stoloniferous species were not clearly distinct from

those of rhizomatous species The second and the

third response groups contained both stoloniferous

(G fluitans and R repens) and rhizomatous species

(C divisa E palustris E repens J articulatus and J

gerardii) Thus contrary to the predictions that clonal

Plant Ecol

123

architecture should be more responsive to defoliation

in stoloniferous than in rhizomatous species the

response of clonal architecture to defoliation was not

constrained by the type of connections Other devel-

opmental constraints may be implied in architectural

responses to defoliation For instance branching

pattern is related to the number and activity of

axillary meristems (Huber and During 2001) which

may play a key role in architectural response to

defoliation (Briske 1996) In monopodial species

(eg T fragiferum and T repens) the number of

connections is constrained by the limited number of

meristems available per ramet (Huber and During

2001) As a result they are likely to be more sensitive

to defoliation (Huber et al 1999) Indeed the trajec-

tories of response of the two studied Trifolium species

did show a great decrease of the number of connec-

tions compared to the other eight species (sympodial

species Klimes and Klimesova 1999)

However the species constitutive of the second (J

articulatus J gerardii and R repens) and the third

(C divisa E palustris E repens and G fluitans)

response groups can be linked neither by their

phylogenetic nor by their developmental features

Our results thus demonstrate that architectural

response to defoliation can neither be predicted on

the sole basis of the type of clonal connection

(stolons or rhizomes) nor by the phylogenetic and

developmental features

Conclusion

Species responses to defoliation were either the

stability of clonal architecture or the decreased

investment in the production of connections and a

lower clonal propagation Gross et al (2007) showed

that low lateral spread was a constitutive trait of

species tolerant to clipping Such growth forms can

be interpreted as the expression of the lsquoconsolidation

strategyrsquo (as defined by de Kroon and Schieving

1990) characterised by short and little plastic con-

nections In grazed pastures where the risk of

defoliation is high plants with short propagation

(either constitutive or induced by defoliation) and

capable of producing a dense population of ramets

when defoliated are very likely to be more compet-

itive than plants that allocate energy in the production

of long connections However small variations of

architectural traits may have great consequences on

spatial distribution of ramets within a genet and

consequently on genet performance resource acqui-

sition and competition (Huber et al 1999) Such

parameters are likely to be modified by recurrent

defoliation that occurs in grazed pastures The four

architectural patterns of response to defoliation

distinguished during the present short-term experi-

ment are very likely to constrain competitive ability

and hence the relative species abundances along the

grazing gradient

Acknowledgements We thank Veronique Sauziere

Christine Tyrant Thierry Fontaine and Fouad Nassur for

their technical support to carry out this experiment This

manuscript is part of the ACI-ECCO ECOGER lsquoDIVHERBrsquo

program and of the GDR CNRS 2574 lsquoTRAITSrsquo

References

Belsky AJ (1986) Does herbivory benefit plants A review of

the evidence Am Nat 127870ndash892 doi101086284531

Briske DD (1996) Strategies of plant survival in grazed sys-

tems a functional interpretation In Hodgson J Illius AW

(eds) The ecology and management of grazing systems

CAB International Wallingford pp 37ndash67

Brown B Allen T (1989) The importance of scale in evaluating

herbivory impacts Oikos 54189ndash194 doi102307

3565265

de Kroon H Schieving F (1990) Resource partitioning in relation

to clonal growth strategy In van Groenendael J de Kroon

H (eds) Clonal growth in plants regulation and function

SPB Academic Publishing The Hague pp 113ndash130

Del-Val E Crawley MJ (2005) Are grazing increaser species

better tolerators than decreasers An experimental

assessment of defoliation tolerance in eight British

grassland species J Ecol 931005ndash1016 doi101111

j1365-2745200501011x

Doledec S Chessel D (1991) Recent development in linear

ordination methods for environmental sciences Adv Ecol

1133ndash155

Dong M de Kroon H (1994) Plasticity in morphology and

biomass allocation in Cynodon dactylon a grass species

forming stolons and rhizomes Oikos 7099ndash106 doi

1023073545704

Dong M Pierdominici MJ (1995) Morphology and growth of

stolons and rhizomes in three clonal grasses as affected

by different light supply Vegetatio 11625ndash32

Gardner SN Mangel M (1997) When can a clonal organism

escape senescence Am Nat 150462ndash490 doi101086

286076

Gross N Suding KN Lavorel S (2007) Leaf dry matter content

and lateral spread predict response to land use change for

six subalpine grassland species J Veg Sci 18289ndash300

doi1016581100-9233(2007)18[289LDMCAL]20CO2

Plant Ecol

123

for Trifolium species weaker for C divisa E palus-

tris E repens and G fluitans and was close to zero

for Juncaceae and R repens The trajectory of

A stolonifera along this axis had a singular shape

with a great increase between control and mid-

frequency defoliation and a decrease between mid-

frequency and high-frequency defoliation (Fig 3b)

The HAC based on architectural responses to

defoliation resulted in four groups of species The

first group consisted of both Trifolium species

J articulatus J gerardii and R repens formed the

second group while C divisa E palustris E repens

and G fluitans constituted the third group Finally the

last group corresponded to A stolonifera due to the

singular shape of its trajectory of response (Fig 4)

Discussion

Consequences of defoliation on clonal

performance

All species showed tolerance to defoliation as 311 out

of 330 genets survived and developed even when

severely clipped

Biomass allocation to ramets was not affected by

the treatment except for one species This indicates

that genets were able to sustain damaged ramets and

even to compensate for aboveground biomass

removal caused by defoliation While the design did

not allow to identify the mechanisms involved

B

Tfra Trep

Asto

Rrep

ErepEpal

Gflu

C

M

H

C

M

H

C

M

H

CMH

CM

H

C

M

H

C

M

H

C

M

HC

M

H

C

M

H

Cdiv

Jart Jger

NbCon

11 -1

-1

A

72 2F

-037-11 13

046Tfra Trep

Asto

Rrep

ErepEpal

Gflu

C

M

H

C

M

H

C

M

H

CMH

CM

H

C

M

H

C

M

H

C

M

HC

M

H

C

M

H

Cdiv

Jart Jger

NbCon

LConDist

11 -1

-1

F1 69

-037-11 13

046

Fig 3 Multivariate responses of clonal architecture to defo-

liation for each species Traits values have been centred per

species (within-species PCA) and compared between treat-

ments (between-treatment PCA) The factorial plan is thus the

same for all species a Correlation circle of architectural traits

in the factorial plan 1ndash2 of between-treatment PCA bTrajectories of multivariate responses to the three treatments

for each species in the factorial plan 1ndash2 Each point represents

the barycentre of all replicates of a defoliation treatment

Points C Control treatment M mid-frequency defoliation

treatment H high-frequency defoliation treatment Dist mean

inter-ramet distance LCon mean length of connections

NbCon number of connections See Table 2 for the meaning

of species abbreviations

Tfra

Trep

Rrep

Gflu

Asto

Cdiv

Epal

Erep

Jart

Jger

Fig 4 Dendrogram of species resulting from HAC according

to the multivariate responses of clonal architecture to

defoliation Species were clustered into four groups on the

basis of Euclidean distance (Wardrsquos method) See Table 2 for

the meaning of species abbreviations

Plant Ecol

123

Brown and Allen (1989) reported that clipping

treatment may cause the translocation of resources

from belowground organs to support the regrowth of

aboveground tissues

We recorded however a great discrepancy on the

degree of tolerance depending on traits and species

considered The RGR decreased with clipping which

is typical of undercompensation (Strauss and Agra-

wal 1999) and incomplete tolerance (Stowe et al

2000) while together seven out of 10 species

maintained a similar number of ramets in all three

treatments arguing for compensating tolerance to

clipping Previous studies had already reported a

variety of responses to clipping by clonal plants

among species and among environments for a given

species For example ramet number has been

reported to decrease with leaf removal for three

clonal perennials (Hicks and Turkington 2000) to be

unchanged for Leymus chinensis (Wang et al 2004)

either to be unchanged or to increase according to

nutrient availability for Cyperus esculentus (Li et al

2004) and even to increase for five caespitose grasses

(Richards et al 1988) Biomass responses to

defoliation were also shown to vary from undercom-

pensation (Li et al 2004) to overcompensation

(Hicks and Turkington 2000) probably due to both

the differing species studied and the defoliation

treatment applied Interspecific differences in com-

pensatory ability were notably found both in clonal

and non-clonal species (McNaughton and Chapin

1985 Belsky 1986 Del-Val and Crawley 2005)

Response to defoliation has also been shown to vary

within the same species depending on environmental

conditions (Maschinski and Whitham 1989) The lack

of generality in clonal plant responses to clipping

may also originate from the variety of defoliation

treatments used in the different studies Their impact

on plant growth may indeed differ whether they are

applied at a given date (eg Hicks and Turkington

2000 Wang et al 2004) at a given development

stage (eg Richards et al 1988) or several times (eg

Li et al 2004)

Responses of clonal architecture-related traits

to defoliation

Clonal architecture-related traits did not match the

predicted responses Species followed four types of

architectural response to defoliation going from nil

to high variation Trait values never increased with

clipping and defoliation thus led to fewer connections

andor shorter connections and mean inter-ramet

distances Similar results in response to clipping or

grazing have already been observed in other herba-

ceous clonal plants such as Trifolium repens (Hay and

Newton 1996) Acaena magellanica (Moen et al

1999) Trientalis europaea (Piqueras 1999) or Lymus

chinensis (Wang et al 2004)

Furthermore the results obtained here show that

there is no relationship between the multivariate

pattern of architectural response and the degree of

tolerance to defoliation For instance both the first

and the third response groups included together

species for which defoliation induced no change in

the number of ramets (C divisa E repens and T

repens) and species for which defoliation induced a

decreased number of ramets (E palustris G fluitans

and T fragiferum) Compensating tolerance can be

related to various responses of architectural traits For

E repens the production of ramets was not affected

by defoliation despite a decrease of the number of

connections as only a few ramets were produced by

these connections the majority being caespitose

(tussock forming) An alternative strategy was shown

by C divisa T repens and A stolonifera They

maintained the number of ramets unchanged with

clipping by way of the decreased mean inter-ramet

distance whatever the variation of the other traits

Interspecific comparison of the responses

of clonal architecture to defoliation

Previous studies have shown that higher nutrient and

or light supply increased branching intensities of both

stolons and rhizomes whereas the morphology of

stolons (mean connection length and mean internode

length) was more plastic than the morphology of

rhizomes (Dong and de Kroon 1994 Dong and

Pierdominici 1995) According to our study only the

stolons of A stolonifera showed a high degree of

variability with a high amplitude of variation of all

architecture-related traits The responses of other

stoloniferous species were not clearly distinct from

those of rhizomatous species The second and the

third response groups contained both stoloniferous

(G fluitans and R repens) and rhizomatous species

(C divisa E palustris E repens J articulatus and J

gerardii) Thus contrary to the predictions that clonal

Plant Ecol

123

architecture should be more responsive to defoliation

in stoloniferous than in rhizomatous species the

response of clonal architecture to defoliation was not

constrained by the type of connections Other devel-

opmental constraints may be implied in architectural

responses to defoliation For instance branching

pattern is related to the number and activity of

axillary meristems (Huber and During 2001) which

may play a key role in architectural response to

defoliation (Briske 1996) In monopodial species

(eg T fragiferum and T repens) the number of

connections is constrained by the limited number of

meristems available per ramet (Huber and During

2001) As a result they are likely to be more sensitive

to defoliation (Huber et al 1999) Indeed the trajec-

tories of response of the two studied Trifolium species

did show a great decrease of the number of connec-

tions compared to the other eight species (sympodial

species Klimes and Klimesova 1999)

However the species constitutive of the second (J

articulatus J gerardii and R repens) and the third

(C divisa E palustris E repens and G fluitans)

response groups can be linked neither by their

phylogenetic nor by their developmental features

Our results thus demonstrate that architectural

response to defoliation can neither be predicted on

the sole basis of the type of clonal connection

(stolons or rhizomes) nor by the phylogenetic and

developmental features

Conclusion

Species responses to defoliation were either the

stability of clonal architecture or the decreased

investment in the production of connections and a

lower clonal propagation Gross et al (2007) showed

that low lateral spread was a constitutive trait of

species tolerant to clipping Such growth forms can

be interpreted as the expression of the lsquoconsolidation

strategyrsquo (as defined by de Kroon and Schieving

1990) characterised by short and little plastic con-

nections In grazed pastures where the risk of

defoliation is high plants with short propagation

(either constitutive or induced by defoliation) and

capable of producing a dense population of ramets

when defoliated are very likely to be more compet-

itive than plants that allocate energy in the production

of long connections However small variations of

architectural traits may have great consequences on

spatial distribution of ramets within a genet and

consequently on genet performance resource acqui-

sition and competition (Huber et al 1999) Such

parameters are likely to be modified by recurrent

defoliation that occurs in grazed pastures The four

architectural patterns of response to defoliation

distinguished during the present short-term experi-

ment are very likely to constrain competitive ability

and hence the relative species abundances along the

grazing gradient

Acknowledgements We thank Veronique Sauziere

Christine Tyrant Thierry Fontaine and Fouad Nassur for

their technical support to carry out this experiment This

manuscript is part of the ACI-ECCO ECOGER lsquoDIVHERBrsquo

program and of the GDR CNRS 2574 lsquoTRAITSrsquo

References

Belsky AJ (1986) Does herbivory benefit plants A review of

the evidence Am Nat 127870ndash892 doi101086284531

Briske DD (1996) Strategies of plant survival in grazed sys-

tems a functional interpretation In Hodgson J Illius AW

(eds) The ecology and management of grazing systems

CAB International Wallingford pp 37ndash67

Brown B Allen T (1989) The importance of scale in evaluating

herbivory impacts Oikos 54189ndash194 doi102307

3565265

de Kroon H Schieving F (1990) Resource partitioning in relation

to clonal growth strategy In van Groenendael J de Kroon

H (eds) Clonal growth in plants regulation and function

SPB Academic Publishing The Hague pp 113ndash130

Del-Val E Crawley MJ (2005) Are grazing increaser species

better tolerators than decreasers An experimental

assessment of defoliation tolerance in eight British

grassland species J Ecol 931005ndash1016 doi101111

j1365-2745200501011x

Doledec S Chessel D (1991) Recent development in linear

ordination methods for environmental sciences Adv Ecol

1133ndash155

Dong M de Kroon H (1994) Plasticity in morphology and

biomass allocation in Cynodon dactylon a grass species

forming stolons and rhizomes Oikos 7099ndash106 doi

1023073545704

Dong M Pierdominici MJ (1995) Morphology and growth of

stolons and rhizomes in three clonal grasses as affected

by different light supply Vegetatio 11625ndash32

Gardner SN Mangel M (1997) When can a clonal organism

escape senescence Am Nat 150462ndash490 doi101086

286076

Gross N Suding KN Lavorel S (2007) Leaf dry matter content

and lateral spread predict response to land use change for

six subalpine grassland species J Veg Sci 18289ndash300

doi1016581100-9233(2007)18[289LDMCAL]20CO2

Plant Ecol

123

Brown and Allen (1989) reported that clipping

treatment may cause the translocation of resources

from belowground organs to support the regrowth of

aboveground tissues

We recorded however a great discrepancy on the

degree of tolerance depending on traits and species

considered The RGR decreased with clipping which

is typical of undercompensation (Strauss and Agra-

wal 1999) and incomplete tolerance (Stowe et al

2000) while together seven out of 10 species

maintained a similar number of ramets in all three

treatments arguing for compensating tolerance to

clipping Previous studies had already reported a

variety of responses to clipping by clonal plants

among species and among environments for a given

species For example ramet number has been

reported to decrease with leaf removal for three

clonal perennials (Hicks and Turkington 2000) to be

unchanged for Leymus chinensis (Wang et al 2004)

either to be unchanged or to increase according to

nutrient availability for Cyperus esculentus (Li et al

2004) and even to increase for five caespitose grasses

(Richards et al 1988) Biomass responses to

defoliation were also shown to vary from undercom-

pensation (Li et al 2004) to overcompensation

(Hicks and Turkington 2000) probably due to both

the differing species studied and the defoliation

treatment applied Interspecific differences in com-

pensatory ability were notably found both in clonal

and non-clonal species (McNaughton and Chapin

1985 Belsky 1986 Del-Val and Crawley 2005)

Response to defoliation has also been shown to vary

within the same species depending on environmental

conditions (Maschinski and Whitham 1989) The lack

of generality in clonal plant responses to clipping

may also originate from the variety of defoliation

treatments used in the different studies Their impact

on plant growth may indeed differ whether they are

applied at a given date (eg Hicks and Turkington

2000 Wang et al 2004) at a given development

stage (eg Richards et al 1988) or several times (eg

Li et al 2004)

Responses of clonal architecture-related traits

to defoliation

Clonal architecture-related traits did not match the

predicted responses Species followed four types of

architectural response to defoliation going from nil

to high variation Trait values never increased with

clipping and defoliation thus led to fewer connections

andor shorter connections and mean inter-ramet

distances Similar results in response to clipping or

grazing have already been observed in other herba-

ceous clonal plants such as Trifolium repens (Hay and

Newton 1996) Acaena magellanica (Moen et al

1999) Trientalis europaea (Piqueras 1999) or Lymus

chinensis (Wang et al 2004)

Furthermore the results obtained here show that

there is no relationship between the multivariate

pattern of architectural response and the degree of

tolerance to defoliation For instance both the first

and the third response groups included together

species for which defoliation induced no change in

the number of ramets (C divisa E repens and T

repens) and species for which defoliation induced a

decreased number of ramets (E palustris G fluitans

and T fragiferum) Compensating tolerance can be

related to various responses of architectural traits For

E repens the production of ramets was not affected

by defoliation despite a decrease of the number of

connections as only a few ramets were produced by

these connections the majority being caespitose

(tussock forming) An alternative strategy was shown

by C divisa T repens and A stolonifera They

maintained the number of ramets unchanged with

clipping by way of the decreased mean inter-ramet

distance whatever the variation of the other traits

Interspecific comparison of the responses

of clonal architecture to defoliation

Previous studies have shown that higher nutrient and

or light supply increased branching intensities of both

stolons and rhizomes whereas the morphology of

stolons (mean connection length and mean internode

length) was more plastic than the morphology of

rhizomes (Dong and de Kroon 1994 Dong and

Pierdominici 1995) According to our study only the

stolons of A stolonifera showed a high degree of

variability with a high amplitude of variation of all

architecture-related traits The responses of other

stoloniferous species were not clearly distinct from

those of rhizomatous species The second and the

third response groups contained both stoloniferous

(G fluitans and R repens) and rhizomatous species

(C divisa E palustris E repens J articulatus and J

gerardii) Thus contrary to the predictions that clonal

Plant Ecol

123

architecture should be more responsive to defoliation

in stoloniferous than in rhizomatous species the

response of clonal architecture to defoliation was not

constrained by the type of connections Other devel-

opmental constraints may be implied in architectural

responses to defoliation For instance branching

pattern is related to the number and activity of

axillary meristems (Huber and During 2001) which

may play a key role in architectural response to

defoliation (Briske 1996) In monopodial species

(eg T fragiferum and T repens) the number of

connections is constrained by the limited number of

meristems available per ramet (Huber and During

2001) As a result they are likely to be more sensitive

to defoliation (Huber et al 1999) Indeed the trajec-

tories of response of the two studied Trifolium species

did show a great decrease of the number of connec-

tions compared to the other eight species (sympodial

species Klimes and Klimesova 1999)

However the species constitutive of the second (J

articulatus J gerardii and R repens) and the third

(C divisa E palustris E repens and G fluitans)

response groups can be linked neither by their

phylogenetic nor by their developmental features

Our results thus demonstrate that architectural

response to defoliation can neither be predicted on

the sole basis of the type of clonal connection

(stolons or rhizomes) nor by the phylogenetic and

developmental features

Conclusion

Species responses to defoliation were either the

stability of clonal architecture or the decreased

investment in the production of connections and a

lower clonal propagation Gross et al (2007) showed

that low lateral spread was a constitutive trait of

species tolerant to clipping Such growth forms can

be interpreted as the expression of the lsquoconsolidation

strategyrsquo (as defined by de Kroon and Schieving

1990) characterised by short and little plastic con-

nections In grazed pastures where the risk of

defoliation is high plants with short propagation

(either constitutive or induced by defoliation) and

capable of producing a dense population of ramets

when defoliated are very likely to be more compet-

itive than plants that allocate energy in the production

of long connections However small variations of

architectural traits may have great consequences on

spatial distribution of ramets within a genet and

consequently on genet performance resource acqui-

sition and competition (Huber et al 1999) Such

parameters are likely to be modified by recurrent

defoliation that occurs in grazed pastures The four

architectural patterns of response to defoliation

distinguished during the present short-term experi-

ment are very likely to constrain competitive ability

and hence the relative species abundances along the

grazing gradient

Acknowledgements We thank Veronique Sauziere

Christine Tyrant Thierry Fontaine and Fouad Nassur for

their technical support to carry out this experiment This

manuscript is part of the ACI-ECCO ECOGER lsquoDIVHERBrsquo

program and of the GDR CNRS 2574 lsquoTRAITSrsquo

References

Belsky AJ (1986) Does herbivory benefit plants A review of

the evidence Am Nat 127870ndash892 doi101086284531

Briske DD (1996) Strategies of plant survival in grazed sys-

tems a functional interpretation In Hodgson J Illius AW

(eds) The ecology and management of grazing systems

CAB International Wallingford pp 37ndash67

Brown B Allen T (1989) The importance of scale in evaluating

herbivory impacts Oikos 54189ndash194 doi102307

3565265

de Kroon H Schieving F (1990) Resource partitioning in relation

to clonal growth strategy In van Groenendael J de Kroon

H (eds) Clonal growth in plants regulation and function

SPB Academic Publishing The Hague pp 113ndash130

Del-Val E Crawley MJ (2005) Are grazing increaser species

better tolerators than decreasers An experimental

assessment of defoliation tolerance in eight British

grassland species J Ecol 931005ndash1016 doi101111

j1365-2745200501011x

Doledec S Chessel D (1991) Recent development in linear

ordination methods for environmental sciences Adv Ecol

1133ndash155

Dong M de Kroon H (1994) Plasticity in morphology and

biomass allocation in Cynodon dactylon a grass species

forming stolons and rhizomes Oikos 7099ndash106 doi

1023073545704

Dong M Pierdominici MJ (1995) Morphology and growth of

stolons and rhizomes in three clonal grasses as affected

by different light supply Vegetatio 11625ndash32

Gardner SN Mangel M (1997) When can a clonal organism

escape senescence Am Nat 150462ndash490 doi101086

286076

Gross N Suding KN Lavorel S (2007) Leaf dry matter content

and lateral spread predict response to land use change for

six subalpine grassland species J Veg Sci 18289ndash300

doi1016581100-9233(2007)18[289LDMCAL]20CO2

Plant Ecol

123

architecture should be more responsive to defoliation

in stoloniferous than in rhizomatous species the

response of clonal architecture to defoliation was not

constrained by the type of connections Other devel-

opmental constraints may be implied in architectural

responses to defoliation For instance branching

pattern is related to the number and activity of

axillary meristems (Huber and During 2001) which

may play a key role in architectural response to

defoliation (Briske 1996) In monopodial species

(eg T fragiferum and T repens) the number of

connections is constrained by the limited number of

meristems available per ramet (Huber and During

2001) As a result they are likely to be more sensitive

to defoliation (Huber et al 1999) Indeed the trajec-

tories of response of the two studied Trifolium species

did show a great decrease of the number of connec-

tions compared to the other eight species (sympodial

species Klimes and Klimesova 1999)

However the species constitutive of the second (J

articulatus J gerardii and R repens) and the third

(C divisa E palustris E repens and G fluitans)

response groups can be linked neither by their

phylogenetic nor by their developmental features

Our results thus demonstrate that architectural

response to defoliation can neither be predicted on

the sole basis of the type of clonal connection

(stolons or rhizomes) nor by the phylogenetic and

developmental features

Conclusion

Species responses to defoliation were either the

stability of clonal architecture or the decreased

investment in the production of connections and a

lower clonal propagation Gross et al (2007) showed

that low lateral spread was a constitutive trait of

species tolerant to clipping Such growth forms can

be interpreted as the expression of the lsquoconsolidation

strategyrsquo (as defined by de Kroon and Schieving

1990) characterised by short and little plastic con-

nections In grazed pastures where the risk of

defoliation is high plants with short propagation

(either constitutive or induced by defoliation) and

capable of producing a dense population of ramets

when defoliated are very likely to be more compet-

itive than plants that allocate energy in the production

of long connections However small variations of

architectural traits may have great consequences on

spatial distribution of ramets within a genet and

consequently on genet performance resource acqui-

sition and competition (Huber et al 1999) Such

parameters are likely to be modified by recurrent

defoliation that occurs in grazed pastures The four

architectural patterns of response to defoliation

distinguished during the present short-term experi-

ment are very likely to constrain competitive ability

and hence the relative species abundances along the

grazing gradient

Acknowledgements We thank Veronique Sauziere

Christine Tyrant Thierry Fontaine and Fouad Nassur for

their technical support to carry out this experiment This

manuscript is part of the ACI-ECCO ECOGER lsquoDIVHERBrsquo

program and of the GDR CNRS 2574 lsquoTRAITSrsquo

References

Belsky AJ (1986) Does herbivory benefit plants A review of

the evidence Am Nat 127870ndash892 doi101086284531

Briske DD (1996) Strategies of plant survival in grazed sys-

tems a functional interpretation In Hodgson J Illius AW

(eds) The ecology and management of grazing systems

CAB International Wallingford pp 37ndash67

Brown B Allen T (1989) The importance of scale in evaluating

herbivory impacts Oikos 54189ndash194 doi102307

3565265

de Kroon H Schieving F (1990) Resource partitioning in relation

to clonal growth strategy In van Groenendael J de Kroon

H (eds) Clonal growth in plants regulation and function

SPB Academic Publishing The Hague pp 113ndash130

Del-Val E Crawley MJ (2005) Are grazing increaser species

better tolerators than decreasers An experimental

assessment of defoliation tolerance in eight British

grassland species J Ecol 931005ndash1016 doi101111

j1365-2745200501011x

Doledec S Chessel D (1991) Recent development in linear

ordination methods for environmental sciences Adv Ecol

1133ndash155

Dong M de Kroon H (1994) Plasticity in morphology and

biomass allocation in Cynodon dactylon a grass species

forming stolons and rhizomes Oikos 7099ndash106 doi

1023073545704

Dong M Pierdominici MJ (1995) Morphology and growth of

stolons and rhizomes in three clonal grasses as affected

by different light supply Vegetatio 11625ndash32

Gardner SN Mangel M (1997) When can a clonal organism

escape senescence Am Nat 150462ndash490 doi101086

286076

Gross N Suding KN Lavorel S (2007) Leaf dry matter content

and lateral spread predict response to land use change for

six subalpine grassland species J Veg Sci 18289ndash300

doi1016581100-9233(2007)18[289LDMCAL]20CO2

Plant Ecol

123

Hay MJM Newton PCD (1996) Effect of severity of defolia-

tion on the viability of reproductive and vegetative

axillary buds of Trifolium repens L Ann Bot (Lond)

78117ndash123 doi101006anbo19960103

Hicks S Turkington R (2000) Compensatory growth of three

herbaceous perennial species the effect of clipping and

nutrient availability Can J Bot 78759ndash767 doi101139

cjb-78-6-759

Huber H During HJ (2001) No long-term costs of meristem

allocation to flowering in stoloniferous Trifolium species

Evol Ecol 14731ndash748 doi101023A1011643904616

Huber H Lukacs S Watson MA (1999) Spatial structure of

stoloniferous herbs an interplay between structural blue-

print ontogeny and phenotypic plasticity Plant Ecol

141107ndash115 doi101023A1009861521047

Hutchings MJ (1999) Clonal plants as cooperative systems

benefits in heterogeneous environments Plant Species

Biol 141ndash10 doi101046j1442-1984199900001x

Hutchings MJ Bradbury IK (1986) Ecological perspectives on

clonal perennial herbs Bioscience 36178ndash182 doi102307

1310305

Hutchings MJ Wijesinghe D (1997) Patchy habitats division of

labour and growth dividends in clonal plants Trends Ecol

Evol 12390ndash394 doi101016S0169-5347(97)87382-X

Jiasenski M Bazzaz FA (1999) The fallacy of ratios and the

testability of models in biology Oikos 84321ndash326 doi

1023073546729

Kleijn D van Groenendael J (1999) The exploitation of het-

erogeneity by a clonal plant in habitats with contrasting

productivity levels J Ecol 87873ndash884 doi101046

j1365-2745199900406x

Klimes L Klimesova J (1999) CLO-PLA2mdasha database of

clonal plants in central Europe Plant Ecol 1419ndash19 doi

101023A1009801216504

Klimes L Klimesova J Hendricks R van Groenendael J

(1997) Clonal plant architecture a comparative analysis

of form and function In de Kroon H van Groenendael J

(eds) The ecology and evolution of clonal plants Back-

huys Publishers Leiden pp 1ndash29

Legendre P Legendre L (1998) Numerical ecology 2nd

English edn Elsevier Scientific Publishing Company

Amsterdam

Lennartsson T Nilsson P Tuomi J (1998) Induction of over-

compensation in the field gentian Gentianella campestris

Ecology 791061ndash1072

Li B Shibuya T Yogo Y Hara T (2004) Effect of ramet

clipping and nutrient availability on growth and biomass

allocation of yellow nutsedge Ecol Res 19603ndash612 doi

101111j1440-1703200400685x

Liao M Yu F Song M Zhang S Zhang J Dong M (2003)

Plasticity in RS ratio morphology and fitness-related

traits in response to reciprocal patchiness of light and

nutrients in the stoloniferous herb Glechoma longituba L

Acta Oecol 24231ndash239 doi101016jactao200307001

Loucougaray G Bonis A Bouzille JB (2004) Effect of grazing

by horses andor cattle on the diversity of coastal grass-

lands in western France Biol Conserv 11659ndash71 doi

101016S0006-3207(03)00177-0

MacDonald SE Lieffers VJ (1993) Rhizome plasticity and

clonal foraging of Calamagrostis canadensis in response

to habitat heterogeneity J Ecol 81769ndash776 doi

1023072261674

Macek P Leps J (2003) The effect of environmental hetero-

geneity on clonal behaviour of Prunella vulgaris L Plant

Ecol 16831ndash43 doi101023A1024460525317

Maschinski J Whitham TG (1989) The continuum of plant

responses to herbivory the influence of plant association

nutrient availability and timing Am Nat 1341ndash19 doi

101086284962

McNaughton SJ Chapin FSIII (1985) Effects of phosphorus

nutrition and defoliation on C4 graminoids from the Ser-

engeti plains Ecology 661617ndash1629 doi102307

1938024

Moen J Ingvarsson PK Walton DWH (1999) Estimates of

structural complexity in clonal plant morphology com-

parison of grazed and ungrazed Acaena magellanicarhizomes Can J Bot 77869ndash876 doi101139cjb-77-

6-869

Oborny B Kun A (2002) Fragmentation of clones how does it

influence dispersal and competitive ability Evol Ecol

15319ndash346 doi101023A1016084815108

Owen D (1980) How plants may benefit from the animals that

eat them Oikos 35230ndash235 doi1023073544430

Paige KN Whitham TG (1987) Overcompensation in response

to mammalian herbivory the advantage of being eaten

Am Nat 129407ndash416 doi101086284645

Piqueras J (1999) Herbivory and ramet performance in the

clonal herb Trientalis europara L J Ecol 87450ndash460 doi

101046j1365-2745199900372x

Price EAC Marshall C (1999) Clonal plants and environmental

heterogeneity An introduction to the proceedings Plant

Ecol 1413ndash7 doi101023A1009838300691

Richards JH Mueller RJ Mott MT (1988) Tillering in tussock

grasses in relation to defoliation and apical bud removal

Ann Bot (Lond) 62173ndash179

Stowe KA Marquis RJ Hochwender CG Simms EL (2000) The

evolutionary ecology of tolerance to consumer damage

Annu Rev Ecol Syst 31565ndash595 doi101146annurev

ecolsys311565

Strauss SY Agrawal AA (1999) The ecology and evolution of

plant tolerance to herbivory Trends Ecol Evol 14179ndash

185 doi101016S0169-5347(98)01576-6

Stuefer JF Huber H (1998) Differential effects of light quantity

and spectral light quality on growth morphology and

development of two stoloniferous Potentilla species

Oecologia 1171ndash8 doi101007s004420050624

Suzuki JI Stuefer JF (1999) On the ecological and evolution-

ary significance of storage in clonal plants Plant Species

Biol 1411ndash17 doi101046j1442-1984199900002x

Thioulouse J Chessel D Dodelec S Olivier JM (1997) ADE-4

a multivariate analysis and graphical display software

Stat Comput 775ndash83 doi101023A1018513530268

Wang Z Li L Han X Dong M (2004) Do rhizome severing

and shoot defoliation affect clonal growth of Leymuschinensis at ramet population level Acta Oecol 26255ndash

260 doi101016jactao200408007

Plant Ecol

123